Install any skill in seconds. Free to start, no credit card required.
Get Started Free →用于搜索英文社交平台,包括 Reddit 帖子、Twitter/X 推文和 YouTube 视频。
.claude/skills/opensensenova-sn-search-social-en/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-21 | ✗→✓ | ▲ Improved | 167% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 19% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 53% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -20% | 0% |
API key、token 与 cookie 统一建议写在仓库根目录 .env(参考 .env.example),并由 runtime 或用户在执行前加载为同名环境变量。脚本仍只从环境变量或显式 CLI 参数读取凭证;不要把真实密钥写入 skill payload、报告、日志或提交。
搜索 Reddit、Twitter/X、YouTube 三个英文社交平台。
| 脚本 | 平台 | 用途 | API 密钥 | |------|------|------|---------| | reddit_search.py | Reddit | 帖子和讨论搜索 | 无需 | | twitter_search.py | Twitter/X | 推文搜索 | 需 TIKHUB_TOKEN | | youtube_search.py | YouTube | 视频搜索 | 需 YOUTUBE_API_KEY |
首次运行或脚本提示缺库时,使用本技能的依赖清单安装到当前 Python 环境:
bashpython3 -m pip install -r requirements.txt
不要在脚本内部自动安装依赖。若安装失败、网络不可用或包不可用,停止使用对应脚本并改用网页搜索,说明缺少依赖。
bashpython3 scripts/reddit_search.py <query> [选项]
| 参数 | 说明 | 默认值 | |------|------|--------| | query | 搜索关键词(必填) | — | | --limit, -n | 返回结果数量 | 10 | | --subreddit, -r | 限定子版块(如 python, machinelearning) | — | | --sort | 排序方式:relevance, hot, top, new, comments | relevance | | --time, -t | 时间范围:hour, day, week, month, year, all | all |
bashpython3 scripts/reddit_search.py "machine learning projects" --limit 5 python3 scripts/reddit_search.py "async python" --subreddit python --sort top --time month --limit 5
bashpython3 scripts/twitter_search.py <query> [选项]
| 参数 | 说明 | 默认值 | |------|------|--------| | query | 搜索关键词(必填) | — | | --limit, -n | 返回结果数量 | 10 | | --token | TikHub Token(也可通过 TIKHUB_TOKEN 环境变量设置,必填) | — |
bashpython3 scripts/twitter_search.py "AI agents" --limit 10 python3 scripts/twitter_search.py "LLM" --token your_tikhub_token --limit 5
bashpython3 scripts/youtube_search.py <query> [选项]
| 参数 | 说明 | 默认值 | |------|------|--------| | query | 搜索关键词(必填) | — | | --limit, -n | 返回结果数量 | 10 | | --api-key | YouTube API 密钥(也可通过 YOUTUBE_API_KEY 环境变量设置,必填) | — | | --order | 排序方式:relevance, date, viewCount, rating | relevance |
bashpython3 scripts/youtube_search.py "transformer explained" --limit 5 python3 scripts/youtube_search.py "python tutorial" --order viewCount --limit 10
标准 JSON:{"success": true, "query": "...", "provider": "reddit|twitter|youtube", "items": [...], "error": null}
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-21 | fail→pass | 5,807 | 10,888 | +87% | 1 | 1 | 0% | 923 | 2,461 | +167% | 0 | 0 | — |
case-22 | fail→fail | 14,091 | 10,090 | -28% | 1 | 1 | 0% | 1,700 | 2,815 | +66% | 0 | 0 | — |
case-01 | fail→fail | 7,989 | 6,704 | -16% | 1 | 1 | 0% | 1,231 | 1,356 | +10% | 0 | 0 | — |
case-02 | fail→fail | 5,173 | 9,163 | +77% | 1 | 1 | 0% | 744 | 1,695 | +128% | 0 | 0 | — |
case-03 | fail→fail | 13,160 | 5,402 | -59% | 1 | 1 | 0% | 2,630 | 1,181 | -55% | 0 | 0 | — |
case-04 | fail→pass | 5,773 | 3,922 | -32% | 1 | 1 | 0% | 1,215 | 1,441 | +19% | 0 | 0 | — |
case-05 | fail→pass | 6,843 | 2,336 | -66% | 1 | 1 | 0% | 1,116 | 1,305 | +17% | 0 | 0 | — |
case-06 | fail→pass | 7,452 | 16,865 | +126% | 1 | 1 | 0% | 1,456 | 2,232 | +53% | 0 | 0 | — |
case-07 | fail→pass | 7,755 | 1,908 | -75% | 1 | 1 | 0% | 1,493 | 1,200 | -20% | 0 | 0 | — |
case-08 | fail→fail | 5,006 | 4,584 | -8% | 1 | 1 | 0% | 888 | 1,149 | +29% | 0 | 0 | — |
case-09 | fail→fail | 5,799 | 6,404 | +10% | 1 | 1 | 0% | 273 | 1,374 | +403% | 0 | 0 | — |
case-10 | fail→fail | 13,323 | 5,696 | -57% | 1 | 1 | 0% | 1,481 | 1,128 | -24% | 0 | 0 | — |
case-11 | fail→pass | 9,735 | 3,059 | -69% | 1 | 1 | 0% | 1,810 | 1,335 | -26% | 0 | 0 | — |
case-20 | fail→pass | 4,770 | 2,726 | -43% | 1 | 1 | 0% | 636 | 1,392 | +119% | 0 | 0 | — |
case-12 | pass→pass | 5,066 | 2,015 | -60% | 1 | 1 | 0% | 753 | 1,216 | +61% | 0 | 0 | — |
case-13 | pass→pass | 2,491 | 1,292 | -48% | 1 | 1 | 0% | 325 | 1,054 | +224% | 0 | 0 | — |
case-14 | pass→pass | 13,212 | 2,881 | -78% | 1 | 1 | 0% | 2,068 | 1,451 | -30% | 0 | 0 | — |
case-15 | fail→pass | 8,407 | 2,028 | -76% | 1 | 1 | 0% | 1,255 | 1,253 | -0% | 0 | 0 | — |
case-16 | pass→pass | 2,892 | 12,046 | +317% | 1 | 1 | 0% | 448 | 1,101 | +146% | 0 | 0 | — |
case-17 | pass→pass | 7,297 | 2,019 | -72% | 1 | 1 | 0% | 1,291 | 1,076 | -17% | 0 | 0 | — |
case-18 | fail→pass | 10,181 | 2,440 | -76% | 1 | 1 | 0% | 1,707 | 1,353 | -21% | 0 | 0 | — |
case-19 | fail→fail | 8,211 | 7,774 | -5% | 1 | 1 | 0% | 1,464 | 1,310 | -11% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 15 counted toward the lift figure. The other 7 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +41 percentage points is the difference between those two pass rates over the 15 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.